Global Morphometrics of Barchan Dunes on Mars Revealed by Artificial Intelligence

Barchan dunes are crescentic windblown features that emerge when the sand supply is limited and the wind is approximately unidirectional. The complex morphology of barchan dunes is influenced by of the environmental conditions in which they form. For that reason, barchans are routinely used to investigate arid climates on both Earth and Mars. Motivated to understand how global trends in barchan morphology relate to global and local environmental conditions, we use a convolutional neural network to map and outline barchan dunes on Mars. Our trained model reviewed 137,111 Mars Reconnaissance Orbiter Context Camera (MRO CTX), detecting over a million instances of barchan dunes. From the dunes’ outlines, we derive morphometrics such as dune width, length and horn ratio - and provide an estimate for the dune volume. Our dataset reveals geospatial patterns related to frost coverage and sand supply and is used to infer dune migration directions on a global scale, which are a proxy for the direction of near surface winds.

Speaker: Lior Rubanenko, Stanford